A method for calibrating a simulation model of a heat exchange equipment of a nuclear power plant

Abstract

This invention specifically relates to a parameter calibration method for a simulation model of heat exchange equipment in a nuclear power plant, belonging to the field of parameter calibration for thermal systems in nuclear power plants. The method includes the following steps: determining the mapping relationship between heat exchange equipment parameters and measured parameters by clarifying the sensor measurement point layout of the heat exchange equipment; constructing a physical relationship model of the heat exchange equipment to determine the parameters to be calibrated; obtaining the measured parameters; obtaining estimated values ​​of the parameters to be calibrated based on the physical relationship model of the heat exchange equipment, and combining the EnKF-OSAS algorithm and the measured parameter values ​​to obtain new estimated values ​​of the parameters to be calibrated; substituting the new estimated values ​​of the parameters to be calibrated into the physical relationship model of the heat exchange equipment to verify the parameter calibration effect; and determining whether to return to continue executing the parameter calibration loop based on the parameter calibration effect. This invention significantly improves the calibration accuracy and efficiency of the simulation model of heat exchange equipment in nuclear power plants through a reverse sequential data assimilation path, a two-level joint estimation system, and a distributed parallel computing architecture.

Description

Technical Field

[0001] This invention relates to the field of parameter calibration technology for thermal systems in nuclear power plants, and in particular to a parameter calibration method for simulation models of heat exchange equipment in nuclear power plants based on the EnKF-OSAS algorithm, also known as the inverse sequential-ensemble Kalman filter assimilation algorithm. Background Technology

[0002] Thermal system parameter calibration is a key technology for digital modeling and real-time simulation of nuclear power plants, and its accuracy directly affects the reliability of core power prediction, equipment life assessment, and system operation optimization. As nuclear power units evolve towards intelligence, improving the prediction accuracy of thermal system simulation models has become a critical task, and dynamic parameter calibration methods based on data assimilation are one key technology. However, this technology still faces the following key challenges in practical applications:

[0003] 1. Time Lag in Traditional Ensemble Kalman Filter Parameter Calibration: Ensemble Kalman Filter (EnKF) uses a serial processing flow of "time update → observation update" in the traditional EnKF framework. Under transient conditions in nuclear power plants, parameter estimates can only be passed to the next time step after the observation data of the current time step has been updated. This time structure results in inherent delays in the correction of key parameters such as steam pipe damping systems and heat exchanger heat transfer efficiency, making it difficult to meet the timeliness requirements of ultra-real-time simulation prediction in nuclear safety guidelines.

[0004] 2. Insufficient dynamic adaptability of thermal system simulation models: Existing simulation platforms, such as RELAP5, mostly use fixed parameters or semi-empirical formulas to describe complex heat and mass transfer processes, which can maintain reasonable accuracy under steady-state conditions. However, in long-term simulations involving two-phase flow oscillations and material aging effects exceeding 1000 hours, or in multi-condition switching scenarios such as peak-shaving operation → base load operation, fixed parameter models will accumulate errors due to the following factors: (1) dynamic changes in the roughness of the pipe interior as the operating time increases; (2) nonlinear shift of the turbine characteristic curve under non-design conditions; (3) the temperature or pressure dependence of coolant physical parameters is not corrected in real time.

[0005] 3. Computational efficiency bottleneck in high-dimensional parameter spaces: Nuclear power system-level models typically involve hundreds of parameters to be calibrated, such as material property parameters, boundary condition parameters, and equipment characteristic parameters. Conventional Bayesian filtering methods face the curse of dimensionality. Taking the traditional Markov chain Monte Carlo method as an example, its convergence time for a 100-dimensional parameter space can reach several hours or more, which cannot meet the requirements for online calibration. Existing dimensionality reduction methods, such as principal component analysis, will destroy the physical correlation between parameters in strongly coupled parameter systems such as reactor-steam generator joint models, causing the corrected parameter combinations to lose their practical engineering significance. Summary of the Invention

[0006] This invention aims to address the issues of time lag, insufficient dynamic adaptability of models, and computational efficiency bottlenecks in the parameter calibration of nuclear power plant thermal systems. It proposes a parameter calibration method for simulation models of heat exchange equipment in nuclear power plants based on the EnKF-OSAS algorithm. This method eliminates parameter estimation lag, enhances the dynamic adaptability of models, and breaks through the bottleneck of high-dimensional computational efficiency, thereby ensuring the safe and economical operation of the unit.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] A method for calibrating parameters of a simulation model of heat exchange equipment in a nuclear power plant based on the EnKF-OSAS algorithm includes the following steps:

[0009] S1. By clarifying the arrangement of sensor measuring points in the heat exchange equipment, determine the mapping relationship between the heat exchange equipment parameters and the measured parameters;

[0010] S2. Construct a physical relationship model of the heat exchange equipment and determine the parameters to be calibrated; the physical relationship model of the heat exchange equipment outputs physical parameters based on the input parameters;

[0011] S3. Obtain measurement parameters;

[0012] S4. Based on the physical relationship model of the heat exchange equipment, obtain the estimated value of the parameter to be calibrated, and combine the EnKF-OSAS algorithm and the measured parameter value to obtain a new estimated value of the parameter to be calibrated.

[0013] S5. Substitute the new estimated values ​​of the parameters to be calibrated into the physical relationship model of the heat exchange equipment to verify the parameter calibration effect.

[0014] S6. Based on the parameter calibration effect, decide whether to return to S3 to continue the parameter calibration loop.

[0015] As one possible approach, in S1, the measured parameters include temperature, pressure, flow rate, and heat exchange area.

[0016] As one possible approach, in S2, to adapt to the EnKF-OSAS algorithm, the physical relationship model of the heat exchanger is expressed in state-space form. Specifically, the physical relationship model of the heat exchanger is represented as follows:

[0017]

[0018] Where n represents time t n x n A vector representing the device state variables at time n; x n-1 A vector representing the device state variables at time n-1;

[0019] Hn This refers to the equipment observation operator or matrix used to calculate the predicted value vector of equipment observations;

[0020] θ represents a vector of equipment characteristic parameters, which are also the parameters to be calibrated. These are parameters such as the pipe-side flow resistance coefficient and the shell-side heat transfer coefficient, which affect the accuracy of the model calculation results but cannot be accurately calculated. These parameters do not change over time without considering equipment aging.

[0021] u n-1 y represents the noise of the equipment state change process at time n-1. n v represents the predicted value vector of device observations at time n. n Indicates observation noise;

[0022] According to the modeling requirements, x n and y n The parameters in the code can be converted to each other.

[0023] As one feasible method, S4 obtains estimated values ​​of the parameters to be calibrated based on the physical relationship model of the heat exchanger, and combines the EnKF-OSAS algorithm and the measured parameter values ​​to obtain new estimated values ​​of the parameters to be calibrated, including the following steps:

[0024] S41. Compare the parameter to be calibrated θ with the device status variable x. n Concatenate into an augmented state vector Constructing an augmented state-space system:

[0025]

[0026] S42. Generate initial parameter set: Based on the physical relationship model of the heat exchanger equipment and the prior parameter distribution p(θ), generate M parameter set members. Combined with the initial state set to form an augmented set Generate corresponding process noise samples and observation noise samples

[0027] S43, Smooth Update Phase: Utilizing the current observation y n Update the parameter set from the previous moment;

[0028] S44, Time Update Phase: The dynamic model is re-integrated using the smoothed parameter set, and secondary corrections are made based on observations to generate an analysis set.

[0029] S45. Parameter constraint processing: Applying the updated parameter set θ a,m Apply physical constraints to ensure that they conform to the practical significance of the physical relationship model of heat exchange equipment;

[0030] S46. Extract calibration results: from the analysis set Extract the set mean of parameter estimates from the dataset. Its set variance is used as the calibration result at the current moment.

[0031] As one possible approach, S43, the smooth update phase, includes the following steps:

[0032] S431. Generate a state prediction set using the physical relationship model of heat exchange equipment:

[0033]

[0034] S432, Based on State Forecast Set With device observation operator or matrix H n Estimate the cross-covariance matrix between the parameters and the observations, and calculate the smoothing gain.

[0035] S433, Utilizing current observations y n The parameter set is updated through Kalman grading:

[0036]

[0037] As one possible implementation method, S44, the time update phase, includes the following steps:

[0038] S441. Generate a set of pseudo-forecasts: using the smoothed parameter set {θ} s,m} and state set Reintegral dynamic model:

[0039]

[0040] S442. Calculate the analytical gain: Estimate the analytical gain L based on the cross-covariance between the pseudo-forecast set and the observations. n :

[0041]

[0042] S443. Update Status and Parameters: Combine Observations y n Perform secondary corrections to generate an analysis set:

[0043]

[0044] As one possible approach, in S45, physical constraints include nonnegativity and range limitations.

[0045] As one possible approach, in S5, verifying the effectiveness of parameter calibration involves calculating the mean square error and determining whether to continue calibration based on whether the mean square error converges.

[0046] As one possible approach, the method employs a distributed collaborative simulation architecture, achieving parallel computation of a thousand-dimensional parameter space through parameter partitioning mapping and hardware acceleration technology.

[0047] One possible approach is through hardware acceleration technologies, including GPUs and FPGAs.

[0048] Beneficial technical effects of the present invention:

[0049] The present invention provides a parameter calibration method for a simulation model of heat exchange equipment in a nuclear power plant based on the EnKF-OSAS algorithm, which eliminates parameter estimation lag. By using a reverse "observation update → time update" data assimilation path, parameter correction is achieved 1-3 simulation steps ahead, improving the real-time tracking capability of key parameters such as steam pipe damping coefficient and heat exchange efficiency under transient conditions, and ensuring dynamic matching between core power and turbine load.

[0050] The present invention provides a parameter calibration method for a simulation model of heat exchange equipment in a nuclear power plant based on the EnKF-OSAS algorithm, which enhances the dynamic adaptability of the model. A two-level joint estimation system for parameters and state variables is constructed, integrating parameter evolution equations and an adaptive covariance optimization algorithm to accurately characterize time-varying factors such as material aging and equipment characteristic drift, significantly improving the prediction accuracy of long-term, multi-condition simulations.

[0051] This invention presents a parameter calibration method for simulation models of heat exchanger equipment in nuclear power plants based on the EnKF-OSAS algorithm, which overcomes the bottleneck of high-dimensional computational efficiency. A distributed collaborative simulation architecture is designed, and parallel computation of a thousand-dimensional parameter space is achieved through parameter partitioning mapping and hardware acceleration technology. This improves the system-level model calibration efficiency by an order of magnitude, meeting the real-time requirements of digital twin systems in nuclear power plants.

[0052] This invention presents a parameter calibration method for simulation models of heat exchange equipment in nuclear power plants based on the EnKF-OSAS algorithm, ensuring the safe and economical operation of the unit. Through a closed-loop linkage between dynamic parameter calibration and high-precision simulation prediction, it effectively suppresses pressure or temperature fluctuations in the main steam system, reduces the risk of equipment fatigue damage, and provides core technical support for the intelligent control of nuclear power units. Attached Figure Description

[0053] Figure 1 This is a flowchart of an embodiment of the method for calibrating simulation model parameters of heat exchange equipment in nuclear power plants based on the EnKF-OSAS algorithm of the present invention.

[0054] Figure 2 This is a flowchart illustrating an embodiment of the EnKF-OSAS algorithm. Detailed Implementation

[0055] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs; the terminology used herein in the specification of the application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application.

[0056] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0057] In the following description, numerous specific details are set forth in order to provide a thorough understanding of the invention. However, it will be apparent to those skilled in the art that these specific details are not necessary to practice the invention. In other embodiments, well-known structures, circuits, materials, or methods have not been specifically described in order to avoid obscuring the invention.

[0058] The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings and specific embodiments.

[0059] See Figure 1-2 This embodiment provides a method for calibrating the parameters of a simulation model of a heat exchanger in a nuclear power plant based on the EnKF-OSAS algorithm, including the following steps:

[0060] S1. Clarify the arrangement of sensor measuring points and parameter mapping relationship of the heat exchange equipment.

[0061] S11. Determine the location of measuring points: Based on the structure and function of the heat exchange equipment, determine the placement of sensors such as temperature, pressure, and flow.

[0062] S12. Establish the physical relationship between the parameters to be calibrated and the sensor measurement parameters:

[0063] ΔP = f1(α, w)

[0064] T 出口 =f2(h,A)

[0065] The parameters to be calibrated include the tube-side resistance coefficient α and the shell-side heat transfer coefficient h. The sensor measurement parameters include temperature T, pressure difference ΔP, flow rate w, and heat transfer area A.

[0066] S2. Constructing a physical relationship model and determining the parameters to be calibrated.

[0067] S21. Construction of Dynamic Physical Relationship Model: Establishing State Equation x n =f n-1 (xn-1 ,θ):

[0068]

[0069] Tube-side energy equation: T 管侧,out =T 管侧,in -α·w 管侧 ·Δt

[0070] Shell-side energy equation: T 壳侧,out =T 壳侧,in +h·A·ΔT LMTD ·Δt

[0071] Observation model: Define y n =H n x n +v n H n Extract the corresponding state variables from the sensor, such as y n =[T 管侧,out ,ΔP];

[0072] S22. Determine the parameters to be calibrated: Identify the parameters that need to be calibrated in the physical relationship model, such as the tube-side flow resistance coefficient and the shell-side heat transfer coefficient.

[0073] S3. Obtain measurement parameters

[0074] S31. Data Acquisition: Read data from the data acquisition system. n Sensor data at any time

[0075] S32. Noise Statistics Setting: Set according to sensor accuracy.

[0076] S4, Calibration Parameter Estimation

[0077] S41. Constructing the augmented state vector

[0078] Device status variable x n Concatenate the vector with the parameter θ to be calibrated to form an augmented state vector:

[0079]

[0080] Device status variable x n Such as thermal efficiency, pressure drop, and parameters to be calibrated θ, such as flow resistance coefficient and heat transfer coefficient;

[0081] S42. Initialize collection members

[0082] S421. Prior sampling: From the prior distribution of parameters (e.g., α ~ U(01, 05), h ~ N(500, 50)). 2 Generate M = 100 set members.

[0083] S422. Noise Generation: Sampling Process Noise Observation noise

[0084] S43, Smooth Update Phase

[0085] S431, Forecast Status Set:

[0086]

[0087] S432, Calculate the gain:

[0088] Calculate the covariance of the forecast ensemble and Update the previous time step

[0089]

[0090] S433, Update parameters:

[0091]

[0092] S44, Time Update Phase

[0093] S441. Generate a set of pseudo-forecasts:

[0094]

[0095] S442. Calculate and analyze gain:

[0096]

[0097] S443, Update augmentation status:

[0098]

[0099] S45, Parameter Constraint Processing

[0100] For θ a,m Cutoff: If α < 0.1, set it to 0.1; if h > 1000, set it to 1000.

[0101] S46. Extract calibration results

[0102] S461. Calculate the mean value of the parameters:

[0103] S462. Calculate the uncertainty:

[0104] S5. Verify calibration effect

[0105] S51, Model Prediction: [This will be...] Substitute the equipment physical relationship model to calculate the predicted value.

[0106] S52. Residual Analysis: Calculate the mean square error

[0107] S6, Iterative Decision

[0108] If MSE > threshold or parameter variance does not converge, return to S3 to collect new data and continue calibration; otherwise, terminate the process.

[0109] The parameter calibration method for the simulation model of heat exchange equipment in nuclear power plants based on the EnKF-OSAS algorithm in this embodiment adopts a reverse assimilation path of "observation update → time update" to establish an advanced parameter correction mechanism. By introducing a sliding time window technique, the observation information is pre-fused 1-3 simulation steps in advance. A dynamic parameter-state joint estimation system is adopted to construct a two-level Kalman filter structure. The inner filter updates time-varying parameters such as heat transfer coefficient and flow resistance factor in real time, while the outer filter synchronously corrects state variables such as temperature, pressure and flow rate. An adaptive covariance expansion algorithm is used to suppress filter divergence while preserving the physical correlation of parameters.

[0110] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A method for calibrating parameters of a simulation model of heat exchange equipment in a nuclear power plant, characterized in that, Includes the following steps: S1. By clarifying the arrangement of sensor measuring points in the heat exchange equipment, determine the mapping relationship between the heat exchange equipment parameters and the measured parameters; S2. Construct a physical relationship model of the heat exchange equipment and determine the parameters to be calibrated; the physical relationship model of the heat exchange equipment outputs physical parameters based on the input parameters; S3. Obtain measurement parameters; S4. Based on the physical relationship model of the heat exchange equipment, obtain the estimated value of the parameter to be calibrated, and combine the EnKF-OSAS algorithm and the measured parameter value to obtain a new estimated value of the parameter to be calibrated. S5. Substitute the new estimated values ​​of the parameters to be calibrated into the physical relationship model of the heat exchange equipment to verify the parameter calibration effect. S6. Based on the parameter calibration effect, decide whether to return to S3 to continue the parameter calibration loop.

2. The parameter calibration method for the simulation model of heat exchange equipment in a nuclear power plant according to claim 1, characterized in that, In S1, the measured parameters include temperature, pressure, flow rate, and heat exchange area.

3. The parameter calibration method for the simulation model of heat exchange equipment in a nuclear power plant according to claim 1, characterized in that, In S2, the physical relationship model of the heat exchange equipment is expressed in a state-space form adapted to the EnKF-OSAS algorithm: Where n represents time t n x n A vector representing the device state variables at time n; x n-1 A vector representing the device state variables at time n-1; H n This refers to the equipment observation operator or matrix used to calculate the predicted value vector of equipment observations; θ represents a vector of equipment characteristic parameters, which is also the parameter to be calibrated. It does not change over time without considering equipment aging. u n-1 y represents the noise of the equipment state change process at time n-1. n v represents the predicted value vector of device observations at time n. n This indicates observation noise.

4. The parameter calibration method for the simulation model of heat exchange equipment in a nuclear power plant according to claim 1, characterized in that, S4. Based on the physical relationship model of the heat exchanger, obtain the estimated values ​​of the parameters to be calibrated. Combine the EnKF-OSAS algorithm and the measured parameter values ​​to obtain new estimated values ​​of the parameters to be calibrated. This includes the following steps: S41. Compare the parameter to be calibrated θ with the device status variable x. n Concatenate into an augmented state vector Constructing an augmented state-space system: S42. Generate initial parameter set: Based on the physical relationship model of the heat exchanger equipment and the prior parameter distribution p(θ), generate M parameter set members. Combined with the initial state set to form an augmented set Generate corresponding process noise samples and observation noise samples S43, Smooth Update Phase: Utilizing the current observation y n Update the parameter set from the previous moment; S44, Time Update Phase: The dynamic model is re-integrated using the smoothed parameter set, and secondary corrections are made based on observations to generate an analysis set. S45. Parameter constraint processing: Applying the updated parameter set θ a,m Apply physical constraints to ensure that they conform to the practical significance of the physical relationship model of heat exchange equipment; S46. Extract calibration results: from the analysis set Extract the set mean of parameter estimates from the dataset. Its set variance is used as the calibration result at the current moment.

5. The parameter calibration method for the simulation model of heat exchange equipment in a nuclear power plant according to claim 4, characterized in that, S43, the smooth update phase, includes the following steps: S431. Generate a state prediction set using the physical relationship model of heat exchange equipment: S432, Based on State Forecast Set With device observation operator or matrix H n Estimate the cross-covariance matrix between the parameters and the observations, and calculate the smoothing gain. S433, Utilizing current observations y n The parameter set is updated through Kalman grading: 。 6. The method for calibrating parameters of a simulation model of heat exchange equipment in a nuclear power plant according to claim 4, characterized in that, S44, the time update phase, includes the following steps: S441. Generate a set of pseudo-forecasts: using the smoothed parameter set {θ} s,m } and state set Reintegral dynamic model: S442. Calculate the analytical gain: Estimate the analytical gain L based on the cross-covariance between the pseudo-forecast set and the observations. n : S443. Update Status and Parameters: Combine Observations y n Perform secondary corrections to generate an analysis set: 。 7. The parameter calibration method for the simulation model of heat exchange equipment in a nuclear power plant according to claim 4, characterized in that, In S45, physical constraints include nonnegativity and range limits.

8. The method for calibrating parameters of a simulation model of heat exchange equipment in a nuclear power plant according to claim 1, characterized in that, In S5, verifying the calibration effect of parameters involves calculating the mean square error and determining whether to continue calibration based on whether the mean square error converges.

9. The parameter calibration method for the simulation model of heat exchange equipment in a nuclear power plant according to claim 1, characterized in that, The method adopts a distributed collaborative simulation architecture and achieves parallel computation of a thousand-dimensional parameter space through parameter partitioning mapping and hardware acceleration technology.

10. The parameter calibration method for the simulation model of heat exchange equipment in a nuclear power plant according to claim 9, characterized in that, Hardware acceleration technologies include GPUs and FPGAs.

Images